GADNN: Gender and Age Detector Neural Network

Automatic prediction of age group and gender from face images has shown good attention, due to the various applications in the domain of recommendation systems and human categorization. The large intra-class variations of the face images due to lightening, scale, occlusion, pose, and angles, made the problem challenging in terms of accuracy. The recent work in the domain of age and gender detection shows better detection accuracy using neural network-based approaches. In this work, a deep learning-based algorithm is proposed for the detection of age groups and gender. The usage of age in most cases is required to identify the age group, so the age detection network detects the age from one of the eight age groups. Both the age group and gender detection neural networks are applied to the benchmark datasets of Adience, UKTFace, and a custom-generated dataset of Pakistani faces. The algorithms resulted in an accuracy ranging from 0.91 to 0.99 on the datasets of gender detection while the accuracies of 0.58 to 0.85 on age detection on benchmark datasets.